Weighted Fair Queuing
Weighted Fair Queuing (WFQ) attempts to address some of the failing of
FIFO and Priority queuing by allowing all traffic some access to the interface.
Some characteristics of WFQ include:
■ Queues traffic by flow or conversation.
■ Flows are identified by header information, such as source and destination
IP address, protocol, source and destination ports, and type of
service field value. These are used by a hash algorithm to create a
queue index number.
■ Each interface has a limited number of WFQ queues. Default for most
interfaces is 256; it can be configured from 16–4096.
■ If the number of flows exceeds the number of queues, multiple flows
are placed in the same queue, resulting in less bandwidth per flow.
■ Provides queues for system traffic and RSVP traffic separate from the
WFQ queues.
■ Traffic is weighted by flow, based on IP precedence.
■ WFQ schedules small interactive flows before high-bandwidth flows.
■ Allows lower-weighted flows relatively more bandwidth than higherweighted
conversations.
■ Drops packets from high-volume flows more aggressively than those
of low-volume flows.
■ The hold-queue limit determines how many packets can be held in the
WFQ system before all new packets are dropped (tail drop). The
default is 1000.
■ The congestive discard threshold (CDT) determines how many packets
can be held by the WFQ before it begins dropping packets from highvolume
flows. Packets are dropped from the queues with the most
packets first. Packets are not dropped from low-volume conversations.
The default CDT is 64.
Advantages of WFQ include simplicity of configuration, support on most
platforms and IOS versions, allowing some bandwidth to all traffic, and
dropping more packets from higher-bandwidth flows.
Disadvantages of WFQ include lack of support for tunneling or encryption,
lack of manual control over traffic classification, lack of minimum or fixed
bandwidth guarantees, and chance of multiple flows placed in the same
queue.